Agentset Reranker Leaderboard

Agentset's reranker leaderboard compares reranking models in a RAG pipeline across six datasets using GPT-5 pairwise judgments and ELO ratings.

12rows
eloprimary metric
2026-05-06sampled

Metadata

ID
agentset_reranker_leaderboard
Category
Retrieval
Release
2025-11-07

Metrics

ELO, Win Rate

Latest Results

Rows are parsed from the public README leaderboard. The benchmark compares reranking models across six retrieval datasets using GPT-5 pairwise judgments and ELO ratings.

Rank Subject ELO Model Match Provenance Sampled
1 Zerank 2 1638 — Imported 2026-05-06
2 Cohere Rerank 4 Pro 1629 — Imported 2026-05-06
3 Zerank 1 1573 — Imported 2026-05-06
4 Voyage AI Rerank 2.5 1544 — Imported 2026-05-06
5 Zerank 1 Small 1539 — Imported 2026-05-06
6 Voyage AI Rerank 2.5 Lite 1520 — Imported 2026-05-06
7 Cohere Rerank 4 Fast 1510 — Imported 2026-05-06
8 Qwen3 Reranker 8B 1473 — Imported 2026-05-06
9 Contextual AI Rerank v2 Instruct 1469 — Imported 2026-05-06
10 Cohere Rerank 3.5 1451 — Imported 2026-05-06
11 BAAI/BGE Reranker v2 M3 1327 — Imported 2026-05-06
12 Jina Reranker v2 Base Multilingual 1327 — Imported 2026-05-06